A Seven-Year-Old Chip Startup Just Raised $468 Million to Fix the AI Memory Shortage

Key points

• Kepler Computing, founded in 2018, emerged from stealth this week after raising $468 million from backers including Intel Capital, AMD Ventures, and Bill Gates.
• The US Department of Commerce committed up to $245 million in July to help Kepler build high-performance AI memory technology inside the United States.
• Kepler's approach skips extreme ultraviolet lithography, an expensive chip-printing technique, and instead uses new materials and 3D stacking to pack more memory into existing factories.
• The startup plans to ship its first sample chips later this year, scale production in Singapore in 2026, and begin US chip production in 2028.
• Kepler has tested its process on roughly 2,000 wafers so far, a small number compared to the millions a full-scale factory would need.

When your phone slows down, your streaming buffers, or an AI chatbot takes a second too long to answer, memory is often the culprit. Not the kind you lose with age. The physical chips inside computers that hold data while processors work through it. Right now the world does not have enough of the fast, dense kind that AI systems need, and building more from scratch costs tens of billions of dollars and takes years.

A startup called Kepler Computing thinks it has a faster, cheaper path.

What is Kepler actually doing?

Kepler is not building new chip factories. It is retooling existing ones to produce a better grade of memory. The company came out of stealth this week after seven years of quiet development, as first reported by Wired AI.

The target is a type of memory called high-bandwidth memory, or HBM. Think of HBM as a fast lane between a computer's processor and its data. AI models, the software systems behind tools like ChatGPT, need enormous amounts of data delivered at very high speed. HBM does that job, but demand has exploded while supply has crawled.

Typically, chipmakers use a process called extreme ultraviolet lithography, or EUV, to print incredibly fine patterns onto silicon, fitting more circuitry into less space. EUV machines cost hundreds of millions of dollars each. Kepler says it can achieve similar density without EUV at all, using a combination of 3D stacking (layering memory chips on top of each other like a tall sandwich rather than spreading them flat) and a new composite material built around a property called ferroelectricity, which allows data to be read and written at lower power than conventional methods.

The team went through 35 different versions of that composite material before landing on one they believed would work at scale.

Who is backing this, and how much?

GlobalFoundries, a major chip manufacturer with a facility in Singapore and another in Burlington, Vermont, is both a financial backer and the factory partner where Kepler runs its tests. Kepler says it converted one of those facilities into a next-generation memory fab in eight months. A normal conversion takes about 24 months.

Should investors and ordinary people get excited yet?

Not too fast. Kepler has processed roughly 2,000 wafers, the disc-shaped sheets of silicon from which chips are cut. A commercial factory churns through millions. Getting consistent, high-quality results at that scale is where many promising chip startups have stalled.

There is also a practical materials problem. Kepler's composite contains iron, a well-known contaminant in chip production. Iron particles can ruin an entire batch of wafers if they escape into the production flow. GlobalFoundries says Kepler's process has to run on dedicated equipment, fully separated from other production lines.

In plain terms: the science appears to work in the lab. The hard part is making it work reliably, cheaply, and in quantity.

For ordinary people, the payoff, if Kepler delivers, would be AI tools that run faster and cost less to operate, because the companies running them spend less on specialised memory. That is a real benefit, but 2028 is the earliest realistic date for US production.

What happens next?

Kepler plans to ship its first HBM sample chips to potential customers before the end of 2025. Volume production in Singapore is targeted for 2026. US production is pencilled in for 2028, supported by the Department of Commerce funding.

The honest takeaway: watch whether those sample chips land on schedule and whether big chip buyers place orders after testing them. That is the real signal that Kepler's approach works outside a controlled lab.